DOES FOREIGN AID PROMOTE TRADE? EVIDENCE FROM SOME ... - Elpelpjournal.eu/wp-content/uploads/2018/03/1-1-2.pdf · to realize the MDGs by the 2015 deadline” therefore trade is vital

European Journal of Contemporary Economics and Management May 2014 Edition Vol.1 No.1

15

DOES FOREIGN AID PROMOTE TRADE?

EVIDENCE FROM SOME SELECTED AFRICAN

COUNTRIES

Paul Ojeaga, PhD Industrial Economics Bergamo Italy Research Scholar Bentley University Boston 2011 to 2012

Note This Paper is part of my PhD Dissertation deposited on AISBERG

repository Bergamo University Italy I also acknowledge the useful incites

from members of faculty of Economics and Mathematics. The views

expressed in the paper are solely those of the Author.

Doi: 10.19044/elp.v1no1a2 URL:http://dx.doi.org/10.19044/elp.v1no1a2

Abstract

Will aid given to specific sector promote growth? Will bilateral aid

be more effective than multilateral aid in export promotion? This paper fills a

gap in the literature by studying the implications of aid channeled

specifically to trade and export oriented growth. Many African Countries

look towards increase in trade driven growth as a means of improving living

standards and boosting growth of their economies. Aid given to trade in

desperately poor countries can be of tremendous advantage to such countries.

We investigate some peculiar components of temporal self limiting aid (often

referred to as development assistance) to sectors that can affect trade in

developing countries. Aid to four sectors was found to have significant

impact on trade although the presence of natural resources tends to reduce

the effectiveness of aid in promoting trade. Institutions and government

economic policies were also found to be weak in the African countries in our

sample reducing aid overall effect on trade.

Keywords: Foreign Aid, Government Economic Policy, Institutional

Quality, Trade

JEL: F13, F16, O24.

1.0 Introduction

The International Trade Centre (ITC) taskforce report 2010/13 states

that global markets and export oriented growth are an effective way of

alleviating poverty, improving livelihood and supporting entrepreneurship in

a sustainable way. The Millennium Development Goals (MDGs) Gap

taskforce report 2010 have also confirmed that “trade is a useful mechanism

http://dx.doi.org/10.19044/elp.v1no1a2


16

to realize the MDGs by the 2015 deadline” therefore trade is vital in driving

growth. Studying aid-trade dynamics in this paper13 provides an insight into

the possible effects that aid can have on trade with particular emphasis on

Africa.

Exports (as a measure of trade) can have a strong effect on the living

conditions of people in advanced countries where strong trading capabilities

have already been attained and in developing countries that are emerging

from economic and political disruptions who wish to establish strong trade

ties (see ITC Trade Report 2010/2013 and DFID (Department for

international Development) Strategy Report Working Paper on Economic

Growth 2008/2013). Exports also has the capability to increase employment,

improve earning power, raise government revenue to provide more services

to its population, provide economic empowerment for the poor through

commerce and deal with environmental and climatic problems, if the gains

accruing from exports are used for common good (see International Trade

Centre (ITC) Taskforce Report 2010/13 for details).

Africa’s per capita GDP is also significantly low and that is why it

remains the World poorest continent (see World Bank Statistics 2012). The

African economy requires a strong industrial effort to drive it out of its

current economic doldrums. The richest countries in Africa based on World

Bank 2009 statistics are South Africa and Egypt, (measured by their

purchasing power parity) with South Africa’s GDP being around $488.6

billion as of 2009, see Economic Watch “An African Economy Overview”

(2010), however many African countries remain extremely poor. Foreign aid

given to these poor countries can have strong effects on the economy of

these countries (particularly in sub Saharan Africa), especially in

circumstances where it consist of a significant percentage of their national

budget or gross domestic product. If the reason behind giving foreign aid is

purely altruistic, then foreign aid can have a positive impact on the recipient

country’s development, if well utilized.

The aims of this study is to identify what components of foreign aid

(channeled to trade) is useful in promoting productivity and increasing trade

in developing countries by dividing aid into sectors, and secondly to

determine what the impact of government economic policy and institutional

quality is on aid effectiveness in Africa? Previous papers have estimated

reduced forms equations of trade aid dynamics they find that country specific

economic and social variations and its proximity or distance to both local and

foreign markets have a significant effect on trade (particularly exports)” see

Morrissey, Osei and Lloyd (2004). In this paper it is assumed that aid is 13 I wish to express my thanks to Bergamo University Italy for funding the course of this research and for the guidance and feedback obtained in writing this paper.


17

endogenous. The reason for this is that aid is likely to suffer from

measurement problems since the data for aid might not capture all aid flows

to developing countries for instance.

The econometric approach we use in the study is instrumental

variable estimation, this allows us to control for endogenity. Aid to sectors

was found to have little or no significant effect on trade using OLS, and it

has a somewhat increased effect in sectors using two stage least squares.

Lots of literatures have argued that aid is useful, while others have suggested

that giving aid does not help developing countries achieve economic growth.

Some papers e.g. that of Jempa (1991)14 examine vast literature on foreign

aid and finds that foreign aid has the tendency to overshadow private

savings, contribute to consumption spending and has no significant impact

on a country’s growth. Others like Boone (1994, 1996) find out that aid has

no effect on investment.

Burnside and Dollar (1997) also find that aid is only beneficial to

countries that have good policies in place. Other authors have found some

similar inconclusive results regarding aid and growth (see Bourguignon and

Sundberg, 2008; Douclouliagos and Paldam, 2007). There is also,

conflicting evidence that aid may have a positive impact on growth

(Gormanee et al, 2003; McPherson and Rakowski, 2001). In addition to

endogeneity, Svensson (2000) finds that disaggregating aid into sectors is a

more promising route in trying to identify the effects that aid can have on a

developing country. Clemens et al (2004) uses sectoral aid and finds a

positive short-run effect on economic growth and that institutional factors

may impact the effectiveness of aid. See Renzio (2006) or Jensen (2008) for

a review of aid literature. The rest of this paper is divided into five parts, the

theoretical part, data description, some constraints to trade in Africa and

index construction, empirical analysis and conclusion.

2.0 Theory and Methodology

The theoretical model we present suggest that some specifically

channeled aid can influence exports, the simple export demand model as

developed by Fontagne et al (2002)15, used by Morrissey et al (2004) and

Cali and Velde (2009), shows the possible effect that aid can have on trade.

If we assume a situation where each country produces one good,

differentiated from the others as a result of its place of production (origin),

with the supply of each good constant and consumers having identical and

14 Jempa, C. (1991) suggests that foreign aid in most cases does not contribute in a significant manner to a country’s economic growth. 15 The model first developed by Fontague et al (2002) describes the role of country specific effect particularly infrastructure in determining trade cost. We extend this model to the private sector and show how aid will affect factors of production such as cost of capital, labor and cost of developing a suitable environment for trade.


18

homothetic choices which is represented by a constant elasticity of

substitution (CES) utility function. The overall utility function of individuals

in a given country k can be defined as a sum of the utility function of

individuals in country i. we represent this in equation 1 below as

(𝟏. ) 𝑼𝒌 = (∑ 𝝋𝒊 𝟏

𝝈

𝑵𝒊=𝟏 𝑪𝒊𝒌

𝝈−𝟏

𝝈

)𝝈

𝝈−𝟏

Where 𝜎 = elasticity of substitution of all goods and services, this can be

defined as the share of goods and services from country i expended upon by

individuals in country k,

C𝑖𝑘 = the value of consumption of the goods produced in country i by

individuals in country k, φ𝑖 = share of goods produced by country i,

expended on by individuals in country k,

where k ϵ [1, N] this subject to the budget constraint that the value of goods

and services consumed by individuals in country k, needs to be equal to the

total national income of country k. The income of country K is given as

shown in equation 2 below as

(𝟐. ) 𝒚𝒌 = ∑ 𝑪𝒊𝒌 𝑷𝒊𝒌 𝑵𝒊=𝟏

Where P𝑖𝑘 =is the price in importer country k of goods produced in

exporting country i

P𝑖 is expressed as the supply price of the exporter i

Then therefore Pik =Pi τik where τik ≥ 1 which is the exporters price times

the cost of transaction

This will capture all types of trade related transaction cost for example

potential tariffs and import taxes, likely administrative cost of trade, ,

transportation to ports and other local and international market destinations

etc see Cali and Velde (2009) for further discussion. After maximizing eqn.1

subject to the budget constraint in eqn. 2 the real consumption Cik with

respect to import of goods from country i by country k is given below as in

equation 3 below

(𝟑. ) 𝐂𝐢𝐤 =𝝋𝒊 𝒀𝒊

𝝉𝒊 𝑷𝒊 (

𝝋𝒊 𝒀𝒊

𝝉𝒊 𝑷𝒊 ) 𝟏−𝝈

The constant elasticity of substitution can be expressed as the likely

trade cost in exporting to country k this can be defined as an index of how far

in terms of distance, that country k is to country i, given by the distance

factor. The distance factor can be defined as how remote goods from country

i are from the market of country k. We express it in terms of remoteness in

eqn. 4

(𝟒. ) 𝐑𝒊 =(∑ 𝛗𝒊 𝛕𝒊 𝟏−𝝈

𝐏𝒊

𝟏−𝝈

𝑵𝒊=𝟏 )

𝟏

𝟏−𝝈

country’s k income can be expressed as y𝑘 = P𝑘 Q𝑘 . which is the price

multiplied by quantity consumed. The total export from country i to country


19

k can be given as X𝑖𝑘

expressed below in terms of the exporting price and

the total consumption in k of country i goods and services.

(5.)

𝐗𝒊𝒌 = 𝐂𝒊𝒌 𝐏𝒊

=𝛗𝒊 𝐘𝒊

𝛕𝒊𝒌 𝝈 (

𝐑

𝐏𝒊 ) 𝝈−𝟏

The total exports from country i can also be expressed as

(𝟔. ) 𝐗𝒊 =𝛗𝒊

𝐩𝒊 𝝈−𝟏∑

𝐘𝒊 𝐑𝒌 𝝈−𝟏

𝛕𝒊𝒌 𝝈 𝑵

𝒌=𝟏

This indicates that exports from country i will depend significantly on

individual country preferences for goods from i. This will depict how

competitive, attractive and the degree to which goods from country i are in

demand in the international market. Total demand from country i therefore,

will be affected in a negative manner by the cost of carrying out trade

transaction in i, since this will affect its final selling price. This will therefore

be displayed by its constant elasticity of substitution σ (CES). If the σ =1

this will mean that the CES is high therefore increases in price of goods from

country i will lead to a significant decrease in exports since buyers will have

to look elsewhere for cheaper goods, therefore change in exports with respect

to price will reduce which we can express as ∂X𝑖

∂p𝑖 < 0.

Incorporating foreign aid into our exports model, it is likely that it

could influence a whole lot of factors that could lead to increase in exports.

Some factors that it could influence are the quality of goods produced by a

country which could lead to product competitiveness in the international

market this will likely increase the country’s share of trade φ𝑖 . Secondly it

could reduce transaction cost in carrying out trade since aid is likely to

improve infrastructure this will make the final price of country i products to

be cheaper. Finally it might also reduce administrative and legal cost since

aid might strengthen both financial and civil institutions so as to increase

access to capital and reduce the bureaucracy of obtaining business permits

and processing exports at ports. We define all these as transaction cost τ𝑖𝑘

and express it below in equation 7 as

(7.) 𝛕𝒊𝒌 = (1+𝐭𝒊𝒌 ) 𝐛𝒊 𝐛𝒌 𝐟 (𝐈𝒊 𝐈𝒌 )𝐝𝒊𝒌

Where τ𝑖𝑘 is the transaction cost of carrying out trade in i relative to

exporting to k. We express b𝑖 and b𝑘 as the cost of processing exports in

country i and k respectively. We assume that transaction cost is a linear

function of distance between i and k therefore country specific infrastructure

should exert a positive effect on transaction cost depending on its state. The

factor d𝑖𝑘 is the trade distance between i and k and therefore a barrier that

should be overcome for trade to take place this could also affect price

significantly, I𝑖 and I𝑘 are the quality of infrastructure in country i and k.

This shows that aid given towards improving trade capacity is likely to

facilitate trade in a positive manner by reducing transaction cost of trade in


20

general. With this we establish that there might indeed exist, a relationship

between trade and aid. This relationship can be expressed as the inverse

function between trade transaction cost and infrastructure. We can express

trade transaction cost to reflect this by writing it as shown in equation 8.

Where I𝑑 is each country’s domestic infrastructure and A4𝐼𝑁𝐹𝑅 is aid

channeled towards infrastructure.

(8.) 𝛕𝒊𝒌 = (𝟏+𝐭𝒊𝒌 ) 𝐛𝒊 (𝐀𝟒𝑻 ) 𝐛𝒌 𝐝𝒊𝒌

(𝐀𝟒𝑰𝑵𝑭𝑹 +𝐈𝑫 )𝒊 +𝐈𝒌

Putting equation 8 into 6 the export becomes

(𝟗. ) 𝐗𝒊 = (𝛗𝒊 (𝐀𝟒𝑷𝑪)) (𝐀𝟒𝑰𝑵𝑭𝑹 +𝐈𝑫 )𝒊

𝝈

𝐏𝒊 𝝈−𝟏 ( 𝐛𝒊 (𝐀𝟒𝑻 ) ) 𝝈

∑ 𝑵𝒋=𝟏

𝐘𝒊 𝐑𝒌 𝝈−𝟏

𝐄𝒊𝒌 𝝈

Where E𝑖𝑘 represents the total cost of trade (in exports) by country i with all

other countries. It is assumed therefore that different types of aid depending

on the sector it is allocated to will likely have a positive effect on exports.

Therefore we express a change in exports with respect to trade as shown in

equation 10 below by including different constraints to trade.

(10.) 𝛛𝐗𝒊

𝛛(𝐀𝟒𝑻)𝒊 =

𝛛𝐗𝒊

𝛛 𝐛𝒌

𝛛𝐛𝒊

𝛛(𝐀𝟒𝑻)𝒊 > 0

A change in exports with respect to aid will depend on how aid given

to boost trade will reduce cost of production within a given country (see Cali

and Velde (2009) for further discussion on how aid can influence export

oriented growth).

Relating this to the private sector, the approach of our model will

depict how aid will affect trade within a country, which will be a

straightforward profit-maximization problem where trade within a country

leads to a situation where firms in the private sector are attempting to

maximize their profits (𝜋𝑖 ). We can express profits ( 𝜋𝑖) as the difference

between total revenue and total cost 𝜋𝑖 = 𝑇𝑅𝑖 − 𝑇𝐶𝑖. In constructing the total

revenue function we will for simplicity assume that firms' quantity choice

does not impact the output price. This will be particularly true for firms in

the export sector as they will be selling at the world price. The total revenue

function for firms operating in sector i can thus be written as the price of

output from sector i (Pi) multiplied by the output level (Xi), 𝑇𝑅𝑖 = 𝑃𝑖𝑋𝑖. So,

the marginal revenue is equal to the price 𝑀𝑅𝑖 = 𝑃𝑖. In sectors of an

economy firm costs are a function of several factors. These include the cost

of labor (w), the cost of capital (v), transportation costs (t), and rent seeking

(r) . The cost of labor is the wage rate per unit of output produced. The cost

of capital can be viewed as the typical rental price of capital but also more

broadly as to include additional factors impacting the cost of obtaining

capital such as access to credit. Transportation costs are a function of both

the distance to market and more importantly the level of infrastructure. For

example, in many developing countries the distance in kilometers to market


21

is considerably less important than the state of the roads that lead there. Rent

seeking represents the cost of dealing with corrupt government officials

imposed on firms. So, the firms total cost function can be written as 𝑇𝐶𝑖 =𝑓𝑖(𝑤, 𝑣, 𝑡, 𝑟)𝑋𝑖.The marginal cost (MC) can be expressed as 𝑓𝑖(𝑤, 𝑣, 𝑡, 𝑟). As

firms increase output we can assume that eventually scarcities will occur and

the marginal cost of production will rise. This can occur because of the

rising cost of labor per unit of output produced and/or because of capital

costs per unit is rising. Eventually, there reaches a point at which

equilibrium occurs in firms. This profit maximization point (𝑋𝑖∗) will

represent the point at which 𝑀𝑅𝑖 = 𝑀𝐶𝑖 , also expressed as 𝑃𝑖 =𝑓𝑖(𝑤, 𝑣, 𝑡, 𝑟).

One of the goals of foreign aid (𝑎𝑖) is to improve conditions for

private sector businesses in developing countries. There are many ways in

which this can occur. Foreign aid can increase education and training of

workers, which would lower the firms labor cost per unit produced. So, the

wage cost per unit produced can be expressed as a negative function of

foreign aid, 𝑤𝑖(𝑎𝑖). Aid may also be used to subsidize equipment/technology

purchases for firms or come in the form of credit extensions which may be

used for capital purchases. Therefore, we can write the cost of capital as a

negative function of foreign aid, 𝑣𝑖(𝑎𝑖) . It is common for both multilateral

and bilateral aid to be used for infrastructure projects (roads, harbors,

airports, etc). These would lower the transportation costs for firms resulting

in the following function where transportation costs are a negative function

of foreign aid, 𝑡𝑖(𝑎𝑖). The flow of foreign aid into a sector may have a

negative side effect; however, by increasing the rent seeking behavior of

government officials since more funds flowing into a sector may result in

corrupt officials seeking higher payout from firms. Therefore, the costs

imposed by rent seeking officials is modeled as a positive function of aid,

𝑟𝑖(𝑎𝑖). With foreign aid included in the model we can rewrite the equilibrium

condition as 𝑃𝑖 = 𝑓𝑖[𝑤𝑖(𝑎𝑖), 𝑣𝑖(𝑎𝑖), 𝑡𝑖(𝑎𝑖)𝑟𝑖(𝑎𝑖)] . We can now examine the impact on the equilibrium condition from a

change in foreign aid. We will assume that foreign aid does not impact

output prices, especially for the export driven sector. Therefore, the

differentiation of this condition with respect to foreign aid is only a

differentiation of the marginal cost function. This can be expressed as

(11.) 𝝏𝒇𝒊

𝝏𝒂𝒊=

𝝏𝒇𝒊

𝝏𝒘𝒊

𝝏𝒘𝒊

𝝏𝒂𝒊+

𝝏𝒇𝒊

𝝏𝒗𝒊

𝝏𝒗𝒊

𝝏𝒂𝒊+

𝝏𝒇𝒊

𝝏𝒕𝒊

𝝏𝒕𝒊

𝝏𝒂𝒊+

𝝏𝒇𝒊

𝝏𝒓𝒊

𝝏𝒓𝒊

𝝏𝒂𝒊

first expression on the right hand side (∂fi

∂wi

∂wi

∂ai≤ 0) represents foreign aid

potentially lowering the cost of labor. They potentially lower cost of capital

from aid is represented as 𝜕𝑓𝑖

𝜕𝑣𝑖

𝜕𝑣𝑖

𝜕𝑎𝑖≤ 0. The potential reduction in transport


22

costs is shown as 𝜕𝑓𝑖

𝜕𝑡𝑖

𝜕𝑡𝑖

𝜕𝑎𝑖≤ 0. The possible rise in rent seeking costs is

the last term on the right hand side which is ∂fi

∂ri

∂ri

∂ai≥ 0. Therefore, the

overall impact of foreign aid is combining three potential cost reduction

factors (w, v, and t) with one potential cost increase (r). Whether or not the

overall sign of 𝜕𝑓𝑖

𝜕𝑎𝑖 is greater or less than zero will depend to a large extent

on the quality of a country’s institutions and on how the foreign aid is

directed. If aid is directed towards more productive uses that lower firm'

labor, capital and/or transport costs then this will help turn the prediction

towards lower marginal costs. If marginal costs of production fall for firms

as a result of foreign aid then output in the sector will increase. In other

words, if ∂fi

∂ai< 0 then

∂Xi

∂ai> 0. We do not use the gravity trade model

because aid is typically between the rich and least developed nations

however we use a partial log equation to depict the effect that aid can have

on trade in developing countries. Therefore our model asserts that exports

will depend on a set of exogenous variables 𝑋𝑖,𝑡 and aid. Our set of

exogenous variables consists of a set of variables that affect trade. The model

we present is the trade model below in equation 12. We extend the model to

sectors and relate the effect of aid to sectors to total trade in a country to

determine the effect that aid to each sector has on trade.

(12.) 𝑬𝒙𝒑𝒐𝒓𝒕𝒔𝒊,𝒕 = 𝜷𝟎 + 𝜷𝟏𝑿𝒊,𝒕 + 𝜷𝟐𝑨𝒊𝒅𝒊,𝒕 + 𝜺𝒊,𝒕

We expect our above model, to yield the following hypotheses which will be

tested in this paper for the export sector

Hypothesis #1.) Aid focused directly on export promotion (extensions of

trade credit, etc) will have a positive impact on exports.

Hypothesis #2.) The positive impact of aid on exports will be reduced if the

country has lower institutional quality (more corruption).

Hypothesis #3.) Aid focused on infrastructure investments will increase

exports.

Hypothesis #4.) Due to conditionality, it is expected that multilateral aid will

suffer from less rent seeking and will be directed more towards lowering

firms’ costs as opposed to bilateral aid. Therefore, the positive impact of aid

on exports should be higher for multilateral aid rather than bilateral aid.

Hypothesis #5.) Aid directed towards the agricultural and educational sector

may or may not increase exports depending on whether the aid is promoting

production for export or for domestic consumption.


23

3.0 Data and Sources

The descriptive statistics of all data used, is presented below (see

Table 1). We use panel data in our study. We obtain data for five African

countries, four in sub Saharan Africa and one in North Africa (i.e. Kenya,

Botswana, Ghana, Cameroon and Egypt) for a period of 39 years 1970 to

2008 although some data are missing.

Dependent Variable

Our dependent variable is exports, we use exports as our measure of

trade, it is the total amount of goods and services exported overseas from a

given country in constant US dollars, it however does not capture domestic

trade which is a major limitation. Data for exports is obtained from World

Bank database. Logarithm of exports is taken because the data on exports is

too noisy therefore this helps to resolve scaling issues. Exports overseas

depicts the exporting country’s capacity to exports and its share of oversea

trade which is often its foreign exchange earning capacity, therefore export is

a vital measure of a country’s international trade.

Description of explanatory variables

Data for aid, gross domestic product (GDP), Population, exchange

rate, trade openness, government spending and inflation was also obtained

from World Bank database. Two different measure of aid is used in this

paper. One is effective aid (pure aid) which consist of grants and grants

component of loans, initially constructed by Chang, Fernandez- Arias, and

Serven (1999) and the other is official aid which we described as distorted

and conditional in nature (distorted because donors often require recipient to

use a sizable amount to import goods from donor countries) it also consists

of grants and loans whose grants component is at least 25% according to

World Bank data. This allows us to determine the difference in their

respective impact on trade. Bilateral and multilateral aid is added up to

obtain what we call total effective aid and total official aid respectively.

Effective aid data was available only for a period of 1975 to 1985 and

official aid data for the period of 1970 to 2008. We intend to compare the

difference of the impact of effective aid from that of official aid on trade and

note the difference between aid without conditionality (effective aid) and aid

with conditionality (official aid) on trade since donor often require recipient

to purchase goods from donor countries as a condition for giving official aid,

this could make official aid to be too stringent thereby limiting its impact on

trade.


24

Table-1 Descriptive Statistics

Variable Observations Mean Std. Dev Min Max

Log of exports 195 3.25 0.48 1.21 4.32

Log of GDP/capita 194 13.35 0.84 11.87 15.79

Natural Resources 195 0.6 0.49 0 3

Exchange Rate 195 1.15 1.92 0.0004 7.03

Landlocked Status 195 1.8 0.4 1 2

Economic Policy 190 -1.25 1.09 -1.80 5.22

Institutional Quality 140 -1.67 1.24 -1.83 1.83

Crude price 195 42.72 21.48 15.93 99.11

Life Expectancy 195 55.04 5.11 44.63 68.41

Health Access 140 73.33 22.62 5 99

Inflation 190 15.05 17.4 -3.21 122.88

Openness 195 70.24 30.56 22.25 157.63

Torture 140 0.59 0.61 0 2

Electoral Self Determination 140 0.99 0.73 0 2

Freedom of movement 140 1.01 0.82 0 2

Political Imprisonment 140 0.87 0.83 0 2

Effective Bilateral Aid 105 2.98 2.43 0.42 15

Effective Multilateral Aid 105 1.41 1.37 0.11 6.4

Total Effective Aid 105 4.39 2.98 0.77 16

Official Multilateral Aid 195 1.62 1.64 0.03 8.28

Total Official Aid 195 5.75 3.91 0.17 18.24

Log of Official aid to Education 144 -6.34 1.74 -14.81 -2.99

Log of Official aid to Agriculture 145 -6.1 1.77 -10.68 -3.46

Log of Official Aid to Infrastructure 145 -4.76 1.3 -10.08 -2.19

Log of Official Aid to Trade Policy 130 -6.96 1.89 -13.28 -3.23

Log of Official aid to industry 144 -6.84 1.85 -13.41 -3.04

School enrollment rate 183 88.94 15.64 55.15 120

Life Expectancy in Years 195 55.04 5.11 2 68.4

Source: Authors compilation (from WDI dataset of the World Bank and other sources)

Effective aid to sectors was not available for individual sectors,

comparing the difference in total aid allocation allows us to know the

difference of the impact of official aid from effective aid on trade, so as draw

conclusion of their impact on trade. Official aid to sectors alone was used to

determine the impact of aid to sectors on trade this will probably affect our

results since we lack data on effective aid to sectors. Data on official aid to

sectors was obtained from The College of William and Mary Williamsburg


25

Virginia aid data base www.aiddata.org , for the period of 1980 to 2008 (29

years) for five the sectors that we wish to consider their effect on trade,

although some years of data are missing. The sectors are trade and business

support services, infrastructure, education agriculture and industry. Country

specific income was represented as GDP per capita which is the average per

capita income of individuals in a country, exchange rate is the average dollar

local currency exchange rate by country this captures fluctuation in the

global economy that are likely to affect trade since the dollar is the global

currency used in international trade. Economic liberalization rate was

captured using the number of phone lines, since businesses are likely to

acquire more phone lines in a liberalized economy than in a highly regulated

one. We use indices to capture the effect of economic policy and institutional

quality on trade in the presence of aid. This method of construction of the

indices is shown in next section. Economic policy is the fluctuations in

government regulatory decisions reflected in its monetary and trade policies.

We capture this using inflation and trade openness variables and develop a

single index for this using principal component analysis (PCA). Investors are

likely to consider how sound and consistent government economic policy

have been overtime in the cause o their future investment in the private

sector of an economy. While institutional quality is reflective of government

attitude and behavior towards governance. We capture these using political

variables. Institutions will also capture a whole host of factors such as

transaction cost involved in running businesses, the cost and time in

acquiring business permits and the quality of infrastructure which will affect

the cost of transportation to both local and foreign markets, since access

roads linking rural agricultural areas to ports will depend on governments

ability to create enabling environment for trade. We obtain data on

institutions from Brigham University political data, we create an index also

for institutions (see next section for index construction). Foreign direct

investment is the inflow of foreign investment to the private business sector

in a country in constant US dollars, school enrollment was used to capture

the level of skill available in the labor market, this was the total school

enrollment of boys and girls between the ages of 0 to 15 years of age,

therefore we expect that this will affect the overall quality of manpower in

countries which could have an effect on output productivity, all variables are

for a period of 1970 to 2008 except otherwise stated.

4.0 Some Constraints to Trade in Africa: Constructing economic and

institutional indices

Business surveys such as World business environment (WBE) report

and World development reports (WBR) of the World Bank of 1999/2000 and

1996/1997 respectively have listed some constraints to foreign direct

http://www.aiddata.org/


26

investment and trade in Africa (See Table 2). They used a sample of 413 and

540 firms respectively in Africa in the two surveys, and respondents were

asked to determine on a four point scale (1= no constraint and 4 = severe

constraint) for the first survey and six point scale (1= no constraint and 6 =

severe constraint) for the second, to depict

Table-2 Constraints on Trade in Africa WBE(1= no constraint WBR(1= no constraint

4= severe constraint) 6= severe constraint)

Corruption 2.80 Taxes and Regulation 4.50

Weak Infrastructure 2.75 Corruption 4.47

Street Crime 2.70 Weak Infrastructure 4.28

Inflation 2.67 Crime 4.25

Financing 2.64 Inflation 4.11

Organized Crime 2.57 Lack of Access to Finance 3.95

Political Instability 2.43 Policy Uncertainty 3.88

Taxes and Regulation 2.24 Cost Uncertainty 3.75

Exchange Rate 2.15 Regulation of Foreign Trade 3.64

Source: World Bank Business Report (2000), also used by Asiedu (2002)

Note: The table above shows the different constraints on trade in Africa using results from

two World Bank surveys

the extent to which some factors constrained business operation in African

countries for each of the reports. As can be seen above in Table 2,

institutional and economic factors rank highest on the list of constraints to

business and trade in Africa. Corruption, weak infrastructure and crime are

the greatest institutional impediments to trade while inflation and financing

are strong economic impediments to trade. Therefore having a good measure

for institutional quality and economic policy as they affect trade is vital in

determining the dynamics that affect trade in Africa. In this paper we group

most of these constraints into institutional and economic factors and reduce

the number of variables by creating an index which captures their effect.

Developing a single variable from a list of variables that have been

identified to be relevant to our topic under study (trade) makes the discussion

of what the effect of institution or policy is on trade to be easier. Most

variables used to capture institutional quality are often political indicators,

they show the direction of a country’s internal governing style and are often

used to rate the reputation of its government and its inclination to good

governance through its affinity for democratic values. Principal component

analysis (PCA) allows us to create a single index for economic policy and

institutional quality, this is a statistical technique used, to derive summary

measures from a set of variables by capturing their variation. The difficulty

most economists face when considering institutions is that they find

numerous indicators for institutions and it becomes difficult to analyze


27

institutions using every single one of them. The variables we use to capture

institutional quality are country specific freedom of movement and electoral

self determination rate. These variables depict a country’s respect for rights

to social assembly and right to self electoral determination. The reason for

using these two variables is that it allows us to capture country specific

freedom of association since this could affect trade if people are prohibited

from doing business because of their opposition to government or

unnecessary threats to life and property. While electoral self determination

rate allows us to capture political stability and the presence of enabling

environment that can promote trade.

Governments also find it difficult to control many economic

indicators, however some of the economic policies that governments float

are captured using indicators that governments try to control, and some

examples of such policies are its monetary, trade and fiscal policies. To

capture these three policies economist use indicators such as inflation, trade

openness and government spending or budget surplus to measure the effect

of these policies on growth. The difficulty arising from using such indicators

is that if one wants to talk about economic policy as an entity it also becomes

virtually impossible or quite cumbersome using more than one indicator to

discuss the effect of government economic policy on the issue under focus.

Due to this difficulty we use these three indicators (inflation, trade openness

and government consumption spending) to develop a single index for

economic policy. In this paper we show the index construction below. We

create the index for institutional quality using the two variables stated above.

They are freedom of movement (freedmove), which is the right to social

association and electoral self determination rate (elecsd) which captures

political stability as stated earlier. We obtain these variables from Brigham

university data set for political indicators developed by international non-

governmental organizations.

Freedom of association was developed by assigning a score of 0 in

cases where it did not exist, 1 in situations where it was interfered with and 2

in cases where it was present. Electoral self determination rate was measured

by assigning a score of 0 in cases where it did not exist, 1 in a case where it

existed but there were some limitations and 2 in a case where citizens have

ability to exercise full political and voting rights. Principal component

analysis uses a weighted average of the underlying variables above to

develop an index for institutional quality using the matrix of eigenvectors

transformation allowing us to obtain an uncorrelated index from a group of

correlated variables. The result of our scatter matrix plot using the two

variables used for generating institutional quality is presented below in fig 1.

Where the variables 1 and 2 are freedom of movement and electoral self

determination rate respectively. The scatter matrix shown below show that


28

electoral self determination and freedom of movement are identically

distributed (see blue dots) and closely correlated. This might not be clear but

a case where variables are not correlated will be explained when developing

the economic policy index subsequently.

Fig.1 Above shows matrix plot the institutional variable

Table 3 Construction of institutional quality index using eigenvectors

Method of construction

(1)

PCA

(2)

PCA

Variables Component 1 Component 2

Electoral self Determination Rate

0.7071

0.7071

Freedom of movement 07071 -0.7071

Note: The above values are generated using eigenvalue transformation. We used the PCA

command “pca elecsd freedmove”to construct the institutional quality index in Table 3

above. The index captures the variation in the two variables allowing us to generate a single

index by adding the individual principal component of the variables. This index obtained

using PCA is uncorrelated with the dependent variable in our regression analysis.

The results of the eigenvectors values is obtained in Table-1 above,

(we show by hand below how this is constructed although “Stata” does it

automatically) using the PCA command “pca elecsd freedmove”. The index

is obtained by adding the two principal components obtained from the

eigenvalue transformation using the two variables alternately as shown

below.

PC1= (0.7071* elecsd) + (0.7071* freedmove) and PC2= (0.7071* elecsd) -

(0.7071* freedmove) where freedmove = freedom of movement and elecsd =

electoral self determination rate. Institutional quality index is given by

Institutional quality = PC1+PC2 Where PC1 and PC2 are principal

components 1 and 2 obtained from our variables.The score plot is shown

below in figure 2 for the two components to depict the variation in our new

elecsd

freedmove

0

1

2

0 1 2

0

1

2

0 1 2


29

index, it shows that there exist sufficient variations among our variables to

capture the effect of institutions.

Fig.2 Above show the score plots for the institutional variable

An index for economic policy was also created using the three

variables inflation, trade openness and government consumption spending

which captures government monetary, trade and fiscal policies respectively.

Inflation is the change in price of goods over time in US dollars, trade

openness is the ratio of exports to imports by country and government

consumption spending is government welfare spending in US dollars, which

displays its fiscal discipline. The scatter matrix plot for the variables used in

the construction of economic policy is shown below below in figure 3. The

results of our scatter plot show that government consumption spending is not

correlated with inflation but only openness (see comparison of the narrow

blue scatter on the left with openness and inflation). We find that inflation

and openness have a stronger correlation with each other. Using a set of

uncorrelated variables could affect the quality of our index since it could

either reduce the variation of the index or over exaggerate its variation

making the index to have a strong negative or strong positive effect leading

to poor conclusions as to the effect of the index in our study. Government

consumption spending was dropped to avoid such problems and we used

only inflation and trade openness in our index construction. Past literature

e.g. Burnside and Dollar (2000) and Easterly (2003) also lend credence to

our assertion since they state that countries can experience growth or trade

increase with poor fiscal conditions (i.e. growing budget deficit) as most

developed countries have for decades.

-2-1

01

Score

s f

or

com

ponent

2

-2 -1 0 1 2Scores for component 1

Score variables (pca)


30

Fig3 Above show the matrix plot for the economic policy variable

The PCA command used to produce the eigenvectors in Table-4 is

“pca openness inflation”

Table 4 Construction of Economic policy index using eigenvectors

Method of construction

(1)

PCA

(2)

PCA

Variables Component 1 Component 2

Openness

0.7071

0.7071

Inflation 07071 -0.7071

Note: The above values are generated using eigenvalue transformation. Using the PCA

below we also show we construct the economic policy index from our matrix of

eigenvectors as follows using the command “pca openness inflation” in the table above.

Economic policy index is generated from our principal component

eigenvector table shown in Table-4 above as PC1= (0.7071*openness) +

(0.7071*inflation) and PC2= (0.7071*openness)-(0.7071*inflation)

Economic policy index is also obtained from summation of the principal

components shown below as Economic policy index = PC1+PC2 where PC1

and PC2 are our principal component, 1 and 2 respectively. The results of the

score plots also shows the correlation between openness and inflation in

figure 4 whereas figure 5 shows all three variables, in figure 5 we observe

that government consumption spending affects

inflation

openness

govconspb

0

50

100

150

0 50 100 150

0

50

100

150

0 50 100 150

0

50

100

0 50 100


31

Fig4 Fig5

Note: Fig 4 and 5 show the score plots for the policy variable

the spread of our score plots in such a manner as to skew our spread more in

the positive direction (compare the spread on the vertical axis of both

figures). This gives some more leverage as to why it was dropped. See

Abeyasekera (2004) and Schlens (2009) for further discussion on how PCA

produce consistent indices.

5.0 Empirical Analysis

Does Aid Attract Trade?

Our empirical model tries to answer, if aid promotes trade? The

argument we present is that many middle income countries receive higher

amounts of aid than low income countries (see USAID 2010 fast facts), this

can be attributed to the fact that there is a higher volume of trade between

middle income countries and developed countries who give aid. Based on

this one might be tempted to say absolutely that it is trade alone that attracts

aid. When one considers cases like that of Rwanda or Kenya that are

particularly poor countries with little or no mineral resources and a low

volume of trade but receive aid, we can state otherwise since there exists

little or no incentive of giving aid to such countries. One can argue from

sound judgment based on reasons for giving aid, that aid is initially altruistic

to poor developing countries. Past studies e.g. Easterly (2003) suggest that

giving aid to a country can establish a close connection between two

countries leading donor country, to search for the presence of minerals and

other country specific endowments in recipient country, on finding a sizeable

deposit of resource this could lead to trade between both countries giving

leverage to the argument that aid could be initially altruistic in nature,

causing aid to attract trade.

-6-4

-20

2

Score

s f

or

com

ponent

2

-2 0 2 4 6Scores for component 1


-10

-50

5

Score

s f

or

com

ponent

2-2 0 2 4 6 8

Scores for component 1



32

Model specification

Hausman specification test was run to choose between using fixed

and random effects model for estimation. Results accept the null hypothesis

that the fixed effects estimator is not biased (p-values are all considerably

lower than .01). The use of instrumental variables approach is because of the

endogeneity of the aid variables. A Hausman-Wu test rejected the null

hypothesis that aid was exogenous, with a p-value of 0.00. Therefore, using

aid as an independent variable could lead to biased results. We then used

fixed effect method of estimation to estimate our equations. We present the

reduced forms of our three versions of aid and trade equations below (for

effective, official and aid to sectors).

The EDA versions of these equations are

(13a) 𝑬𝑫𝑨𝒊,𝒕 = 𝜶𝟎 + 𝜶𝟏𝑿𝒊,𝒕 + 𝜶𝟐𝑰𝒊,𝒕 + 𝐜𝐢 + µ𝒊,𝒕

(13b) 𝑬𝒙𝒑𝒐𝒓𝒕𝒔𝒊,𝒕 = 𝜷𝟎 + 𝜷𝟏𝑿𝒊,𝒕 + 𝜷𝟐𝑬𝑫�̂�𝒊,𝒕 + 𝐜𝐢 + µ𝒊,𝒕

The ODA versions of these equations are

(14a) 𝑶𝑫𝑨𝒊,𝒕 = 𝜶𝟎 + 𝜶𝟏𝑿𝒊,𝒕 + 𝜶𝟐𝑰𝒊,𝒕 + 𝐜𝐢 + µ𝒊,𝒕

(14b) 𝑬𝒙𝒑𝒐𝒓𝒕𝒔𝒊,𝒕 = 𝜷𝟎 + 𝜷𝟏𝑿𝒊,𝒕 + 𝜷𝟐𝑶𝑫�̂�𝒊,𝒕 + 𝐜𝐢 + µ𝒊,𝒕

The sectoral versions of these equations are

(15a) 𝑨𝑰𝑫𝒊,𝒕𝒋

= 𝜶𝟎 + 𝜶𝟏𝑿𝒊,𝒕 + 𝜶𝟐𝑰𝒊,𝒕 + 𝐜𝐢 + µ𝒊,𝒕

(15b) 𝑬𝒙𝒑𝒐𝒓𝒕𝒔𝒊,𝒕 = 𝜷𝟎 + 𝜷𝟏𝑿𝒊,𝒕 + 𝜷𝟐𝑨𝑰𝑫𝒊,𝒕𝒋

+̂ 𝒄𝒊 + µ𝒊,𝒕

The trade (exports) equations are linear specifications, where i is the

index for the countries, t the index for time and trade is the logarithm of per

capita export resulting from trade with other countries, aid (be it EDA,ODA

and aid to sectors) is expressed as the logarithm of aggregate aid allocated

for purposes that can stimulate trade. Our vector of exogenous variables

𝑋𝑖,𝑡 consists of a group of specific variables that affect trade they consist of

government economic policy, average dollar-local currency exchange rates

which capture global shocks that affect trade, institutional quality, school

enrollment rate and GDP per capita. We excluded natural resources as a

variable since we experience negative R-squared with its presence but use it

in interacting aid. We assume that aid is endogenous in the trade equations,

since aid is likely to suffer from measurement problems. Therefore we

employ an instrument 𝐼𝑖,𝑡 for aid to capture the effect of aid in our aid

equation. The dynamics that govern the different types of aid in promoting

trade was found to be complex and different from one another. We find that

with some types of aid, trade was found to depend on additional factors,

since aid was to be used in promotion of trade, for instance with effective


33

and official aid we found that such aid will likely be given to assist trade in

the presence of reasonable foreign direct investment (FDI) and economic

liberalization. With aid to sectors which was in fact official aid to sectors, aid

will depend on some level of economic liberalization which allows for

private ownership and some investment in the private sector but not

necessarily foreign direct investment. Where ci represents time invariant

unobserved effects on trade and µit

represents time varying unobserved

effects on trade. The fixed effect method will produce consistent estimate of

the effect of aid on trade by allowing arbitrary correlation between

unobserved time invariant effects (ci ) and explanatory variables in the trade

equations. The consistency of the FE estimators will depend on following

assumption (a.) The time varying unobserved effects µit

are uncorrelated

with the explanatory variables across all time periods. (b.) There is

significant variation in aid flow over time and (c.) The assumption of strict

exogeneity of explanatory variables is fulfilled. The assumption of strict

exogeneity is going rules out feedback effects from aid to trade and country

specific effects over time.

In some other instance we estimate variants of the trade (exports)

equation using GLS including the variable “interact” the interaction between

aid and policy, aid and institutional quality and aid and natural resources

using interaction variables. The predicted value for the instrumented variable

(sectoral aid) was then interacted with the institutional quality, economic

policy or natural resources. The second equation (with exports as the

dependent variable) then included the predicted aid variable, one of the

interaction variables and the other explanatory variables used in all

regressions.

The interaction versions of these equations are

(16a) 𝑨𝑰𝑫𝒊,𝒕𝒋

= 𝜶𝟎 + 𝜶𝟏𝑿𝒊,𝒕 + 𝜶𝟐𝑰𝒊,𝒕 + 𝒗𝒊 + 𝜺𝒊,𝒕

(16b) 𝑬𝒙𝒑𝒐𝒓𝒕𝒔𝒊,𝒕 = 𝜷𝟎 + 𝜷𝟏𝑿𝒊,𝒕 + 𝜷𝟐𝑨𝑰𝑫𝒊,𝒕�̂�

+ 𝜷𝟑𝒊𝒏𝒕𝒆𝒓𝒂𝒄𝒕𝒊,𝒕 + 𝒗𝒊 + 𝜺𝒊,𝒕

In all government economic policy and income were lagged by one

period. We have a total of two right hand side endogenous variables

(logarithm of trade and aggregate aid) and at least 6 excluded exogenous

variables (logarithm of income, government economic policy, and

institutional quality, exchange rates, market access, foreign direct

investment, economic liberalization rate and school enrollment rate) with

three interaction variables (aid interacts with economic policy, institutional

quality and natural resources).


34

Instrument

Exclusion restriction assumptions are typically theoretical an

instrument that is valid should therefore be correlated with aid in our model

specification but not with trade (exports). One of the most important aspects

of the instrumental variable approach is having a variable (or variables) in

the aid equation which is not included in the trade (export) equation; these

variables are referred to as the “instruments”. We expect that our instruments

should fulfill certain conditions in our case which will be particular for

exports. First, instruments should have a significant impact on the variable

they are predicting, in this case the aid variable. The second condition is that

the instrument should not have an impact on the dependent variable, exports

in the second equation. While often this is tested empirically, Wooldridge

(2010) and others have pointed out that this also needs to be done on the

theoretical level as testing the impact of the instrument on the dependent

variable in the second equation (exports) with a full model could be biased as

the instrumental correction has not been made for the endogenous variable

(aid). We use one instrument “life expectancy” for aid (i.e. for bilateral,

multilateral and aid to sectors for the three sets of equations), so our model is

exactly identified, we expect aid to flow to areas with low life expectancy,

making life expectancy to be positively correlated with aid. Our exclusion

restriction will hold since it is reasonable to state that on the long run low life

expectancy will attract foreign aid but will not promote exports allowing us

to solve our first stage and second stage equations simultaneously.

The exclusion restriction we impose on our trade equation is that life

expectancy is correlated with aid but not with trade, this will hold

econometrically, if the coefficient for aid in our structural equation after

imposing the restriction in our trade equation (where we use life expectancy

as a proxy for aid) tends to that in our reduced form equation and secondly, if

the correlation between the instrument 𝐼𝑖,𝑡 and the error term 𝜀𝑖,𝑡 is

identically equal to zero as shown below in equation 17.

(17) E|𝑰𝒊,𝒕 . 𝒖𝒊|=0 and E|𝑰𝒊,𝒕 . 𝝐𝒊|=0

This then shows that the only way life expectancy is related with trade is

only through aid therefore the instruments 𝐼𝑖,𝑡 is not correlated with the

disturbances ( 𝑢𝑖 and 𝜖𝑖 ) in our model specification and finally, if the

exogenous component of the instrument, (the fitted value of aid) is

uncorrelated with the error term we can therefore identify the variation of the

dependent variable trade (exports) as the slope of the aid coefficient. This

shows that there is sufficient variation (which is non zero) between aid and

our instrument which we represent in the covariance (cov) equation 18

below.

(18) Cov (𝑨𝒊𝒅𝒊,𝒕 . 𝑰𝒊,𝒕) ≠ 0


35

(where Aid can be EDA or ODA, this means that 𝛼2 in not zero). This

implies that exports will vary according to changes in aid in flow to countries

(see Kraay (2008) for further discussion on exclusion restriction). We argue

that our instrument meet the criteria theoretically for our exclusion restriction

to hold, since the behavior of the instrument life expectancy (see the first

stage results and F-tests in Tables 5 to 7), support previous literature e.g.

Heintz (2004) that argue that a good instrument should capture the variation

in the dependent variable and be highly correlated with the endogenous

variable therefore 𝛽2 (our aid coefficient) will no longer biased in our model

specification.

Table 5. First Stage: EDA Regressions Method of Estimation

OLS

Bilateral EDA

OLS

Multilateral EDA

OLS

Total EDA

Life Expectancy 0.36

0.35

0.71

(.12)***

(.11)***

(.20)***

Policy Index -0.06

-0.21

-0.30

(.12)

(.15)

(.23)

Institution Index -0.0004

-0.31

-0.04

(0.35)

(0.17)

(0.34)

Exchange Rate (LCU per $) 0.27

0.23

0.49

(.26)

(.11)**

(.30)

FDI -0.06

-0.06

-3.08

(.89)

(.06)

(.10)**

School Enrollment Rate 0.01

-0.12

-0.11

(.07)

(.03)

(.08)

Liberation Policy -0.42

0.97

1.38

(.88)

(.36)**

(1.03)

GDP per capita -2.67

-0.47

2.07

(1.39)*

(.66)***

(1.67)

F-Test 8,23

10.13

12.90

Chi2 (p-value) 0.00

0.00

0.00

# of observations 73

73

73

R-Squared 22

52

36

Notes: Coefficients listed with standard errors in parentheses. *, ** and *** refers to

significance at the 1%, 5% and 10% levels, respectively. First stage results in Appendix.


36

Table 6. First Stage: ODA Regressions Method of Estimation

OLS

Bilateral EDA

OLS

Multilateral EDA

OLS

Total EDA


0.16

0.44

(.05)***

(.04)***

(.08)***

Policy Index -0.24

-0.35

-0.56

(.15)

(.15)**

(.26)**

Institution Index 0.24

-0.03

0.26

(0.20)

(0.16)

(0.32)

Exchange Rate (LCU per $) -0.42

0.22

-0.18

(.31)

(.13)

(.40)

FDI -0.08

0.03

3.10

(.06)

(.04)

(.09)

School Enrollment Rate 0.02

-0.06

-0.04

(.03)

(.02)***

(.04)

Liberation Policy -0.63

-0.13

-1.75

(.12)***

(.07)*

(1.58)


-0.56

4.47

(.75)***

(.39)

(1.02)***

F-Test 37.28

13.54

29.52

Chi2 (p-value) 0.00

0.00

0.00


131

131

R-Squared 0.53

0.34

0.48




37

Table 7. First Stage:Sectoral Aid Regressions

Aid to

Trade

Aid to

Infrastructure

Aid to

Agriculture

Aid to

Education


0.16

0.13

0.18

(.08)***

(.04)***

(.04)***

(.05)***

Policy Index 0.27

-0.02

-0.02

-0.14

(.24)

(.15)

(.17)

(.32)


0.04

0.37

0.39

(.24)

(.13)***

(.14)***

(.20)*

Exchange Rate (LCU per $) -0.05

-0.01

0.30

-0.23

(.33)

(.13)

(.33)

(.25)

School enrollment -0.02

0.01

-0.04

0.04

(.02)

(.01)***

(.02)

(.02)*

Liberalization policy

-0.11

-0.09

0.12

-0.17

(.12)

(.08)**

(.08)

(.09)**


-1.25

-0.03

-2.40

(.52)

(.42)***

(.57)

(.57)***

F-Test 7.95

18.72

10.25

14.24

Chi2 (p-value) 0.00

0.00

0.00

0.00

# of observation 132

140

140

139

R-Squared 0.21

0.43

0.28

0.25



Results

We use fixed effect regression as stated earlier, since the result of the

Hausman test with p-value 0.00016 (we included this in our results only for

aid to sectors results) suggest that fixed effect estimation is more appropriate

for our model, see Baltagi (2005), Baltagi and Wu (2010) and Wooldridge

(2010) for further discussion. We find that the factors that affect effective aid

are quite different from those of official aid since effective aid is pure aid

devoid of conditionality. Time effect “year” is included to control for

16 This did not hold in some cases with effective and official aid


38

differences in exports from countries in years. This allows us to control, for

production shocks and fluctuations in global demands for exports that is

likely to affect volumes of exports. We present our result for effective aid

and official aid below in Tables 8 to 11. We compare the results of the OLS

estimates with those of the 2SLS. As expected the results of the standard

errors of our 2SLS estimates are larger than those of our OLS estimates for

aid see standard errors in Table 8 and 9 for our regressions of trade (using

log of exports) on effective aid and other factors that affect trade. In Table 8

the OLS estimate for bilateral and multilateral effective aid are 0.06 and 0.10

respectively (see coefficients in table 8 Column 1 and 2) it shows that

bilateral aid contributed 6 percentage points towards trade (with p-value

0.037) while multilateral aid contributed 10 percentage points to trade but

had a stronger effect (with p-value 0.023) on trade. But the estimate is quite

different when we use 2SLS in Table 9 the result of the F-test for excluded

instruments for the 2SLS shows that the instrument life expectancy is valid

and highly correlated with aid (see first stage regression results for effective

aid). The estimated aid effect on trade is now 0.15 and 0.16 (see Table 9

Column 1 and 2) using 2SLS showing that multilateral aid contributed about

1 percentage point more to trade (with p-value 0.03) compared to bilateral

trade with only 15 percent (with p-value 0.07) . This showed that controlling

for endogeneity helps solve the problem of aid measurement, through the

instrumental correction of aid since this could lead to bias in our results.

The results of the regression of trade on official aid are presented in

tables 10 and 11. The result of the OLS regression with estimates for

bilateral and multilateral official aid respectively of 0.06 and 0.05 in Table

10 Column 1 and 2, shows that bilateral aid contributed 6 percentage points

to trade (with p-value of 0.000), while multilateral aid contributed 5

percentage points (with p-value of 0.052) to trade which is 1 percentage

points less than bilateral aid contribution to trade. The results of our 2SLS

estimates where we control for endogeneity are different from our OLS

estimates. The result of the F-test for excluded instruments shows that our

instrument life expectancy is valid and highly correlated with aid (see first

stage official aid regression). The result in Table 11 Columns 1 and 2 shows

that bilateral aid contributes 12 percentage points to trade (with p-value

0.000) while multilateral aid contributes 21 percentage points to trade with (

p-value of 0.000) which is 9 percentage points more than bilateral aid

contribution t trade. This result suggests once again that using 2SLS to

address the issue of endogeneity is important, since aid is likely to suffer

from measurement problems making the OLS results to be biased. Table 12

and 13 present the estimates of the regression of trade on aid to sectors and

factors that affect trade in sectors. The result of our OLS estimates (see Table

12) show that aid to sectors had no effect on trade except for aid to


39

infrastructure that contributed 10 percentage points to trade (with p-value

0.000) while aid to trade policy and business support services, agriculture

and education contributed 2,1 and 3 percentage points to trade respectively

and had no significant effect on trade. The results of our 2SLS are different

from that of the OLS estimates for aid to sectors. Aid had a significant effect

in four sectors, With aid to trade and business support services,

infrastructure, agriculture and education contributing 15, 22, 17 and 16

percentage points respectively to trade using 2SLS (see Table 13 Columns 1,

2, 3 and 4 for aid estimates of 0.15, 0.22, 0.17and 0.16 respectively)

therefore controlling for endogeneity using 2SLS was also relevant in this

case. Aid to industry had no significant effect on trade so we left that out in

our results. The result of our F- test show that our instrument is relevant and

valid since it is highly correlated with aid (see first stage results using

official aid to sectors). Finally GLS was used in estimating our trade

equation with the interactive variables, the three interactive variables

aid*economic policy, aid*institutions and aid*natural resources had reduced

effect on trade showing that these variables reduce aid effectiveness (we

show results in the Appendix-A to C in Tables 14 to 16).Based on the above

results we answer the hypothesis that we posed earlier as follows

Hypothesis #1.) Aid focused directly on export promotion (extensions of

trade credit, etc) was found to be contributing to exporting in a significant

manner. Therefore aid channeled to sectors that could improve output

productivity is likely to be useful in promoting trade.

Hypothesis #2.) Institutions were probably contributing negatively to aid

effectiveness in promoting exports. The interactive variable aid*institutions

had a reduce effect on exporting. It is likely that institutions are weak and not

helping in effective utilization of aid to promote trade.

Hypothesis #3.) Aid focused on infrastructure investments was found to be

contributing to exports in a positive manner. It is likely that aid used in

developing infrastructure will likely create enabling environment that can

promote trade by way of cost reduction in the trade facilitation process.

Hypothesis #4.) Multilateral aid was found to be contributing to exporting in

a more significant manner than bilateral aid. It is likely that the altruistic

nature and good policy requirement conditions associated with multilateral

aid made it more effective in promoting trade than bilateral aid.

Hypothesis #5.) Aid directed towards agriculture and education sector

contributed to exporting significantly. It is likely that aid used in improving

the level of education of the working population as well as modernizing

methods used in cultivation was useful to improving trade.


40

Table 8. Impact of EDA on Exports

Method of Estimation

OLS

(1)

OLS

(2)

OLS

(3)

Bilateral EDA 0.06

-

-

(.03)**

Multilateral EDA -

0.10

-

(.04)**

Total EDA -

-

0.06

(.02)***

Policy Index 0.08

0.07

0.07

(.05)

(.05)

(.05)


0.30

0.28

(.06)***

(.06)***

(.06)***

School enrollment 0.01

0.02

0.02

(.004)***

(.004)***

(.004)***

Exchange rate (LCU per $) 0.08

0.09

0.09

(.04)*

(.04)**

(.04)**

Liberalization policy 0.09

-0.29

0.18

(.10)

(.12)**

(.10)*

FDI 0.06

0.06

0.06

(.02)***

(.02)**

(.02)**

GDP per capita 0.15

0.19

0.22

(.10)

(.10)*

(.10)**

Chi2 (p-value) 0.00

0.00

0.00


73

73

R-Squared 0.72

0.73

0.73




41

Table 9. Impact of EDA on Exports


2SLS

(1)

2SLS

(2)

2SLS

(3)

Bilateral EDA 0.15

-

-

(.09)*

Multilateral EDA -

0.16

-

(.07)**

Total EDA -

-

0.08

(.04)***

Policy Index -0.11

-0.09

-0.10

(.06)**

(.05)*

(.05)*


0.15

0.15

(.06)**

(.04)***

(.05)***


0.02

0.01

(.01)

(.01)**

(.01)

Exchange rate (LCU per $) 0.21

0.21

0.21

(.11)**

(.08)***

(.09)**

Liberalization policy -0.13

-0.04

-0.09

(.17)

(.10)**

(.13)*

FDI 0.01

-0.01

0.001

(.02)***

(.01)**

(.01)**

GDP per capita 0.12

0.61

0.38

(.45)

(.18)***

(.27)

Chi2 (p-value) 0.00

0.00

0.00


73

73

R-Squared 0.28

0.51

0.64




42

Table 10. Impact of ODA on Exports


OLS

(1)

OLS

(2)

OLS

(3)

Bilateral ODA 0.06

-

-

(.02)***

Multilateral ODA -

0.05

-

(.02)*

Total ODA -

-

0.04

(.01)***

Policy Index -0.10

-0.11

-0.10

(.04)

(.05)

(.04)


0.25

0.23

(.04)***

(.04)***

(.04)***

School enrollment 0.01

0.01

0.01

(.003)***

(.003)***

(.003)***

Exchange rate (LCU per $) -0.01

-0.01

-0.01

(.02)

(.02)

(.02)


0.02

0.02

(.02)

(.02)

(.02)

FDI 0.03

0.03

0.03

(.02)*

(.02)**

(.02)*

GDP per capita 0.17

0.12

0.19

(.06)***

(.06)**

(.06)***

Chi2 (p-value) 0.00

0.00

0.00


131

131

R-Squared 0.64

0.61

0.64




43

Table 11. Impact of ODA on Exports


2SLS

(1)

2SLS

(2)

2SLS

(3)

Bilateral ODA 0.12

-

-

(.03)***

Multilateral ODA -

0.21

-

(.06)***

Total ODA -

-

0.07

(.02)***

Policy Index -0.05

-0.01

-0.04

(.07)

(.07)

(.07)


0.16

0.15

(.05)***

(.05)***

(.05)***


0.01

0.04

(.001)

(.004)***

(.003)

Exchange rate (LCU per $) -0.02

-0.12

-0.06

(.07)

(.06)*

(.06)


-0.0002

0.03

(.02)**

(.01)

(.01)*

FDI 0.02

0.02

0.02

(.01)

(.01)*

(.01)*

GDP per capita 0.10

-0.24

-0.02

(.20)

(.13)*

(.16)

Chi2 (p-value) 0.00

0.00

0.00


131

131

R-Squared 0.31

0.36

0.39




44

Table 12 Impact of Sectoral Aid on Exports

Method of Estimation OLS

(1)

OLS

(2)

OLS

(3)

OLS

(4)

Aid to trade 0.2 - - -

(.02)

Aid to infrastructure - 0.10 - -

(.03)***

Aid to agriculture - - 0.01 -

(.02)

Aid to Education - - - 0.03

(.02)

School Enrollment 0.01 0.01 0.01 0.01

(.003)*** (.003)*** (.003)*** (.03)***

Exchange rate -0.02 -0.01 0.0002 -0.01

(.02)*** (.02) (.02) (.02)

Economic policy 0.09 0.11 0.10 0.10

(.05)** (.04)** (.05)** (.04)**

Institutional quality 0.25 0.26 0.29 0.27

(.03)*** (.03)*** (.04)*** (.04)***

Liberalization Policy 0.01 0.01 0.02 0.02

(.20) (.20) (.20) (.20)

GDP per capita 0.07 0.17 0.07 0.08

(.05) (.06)*** (.06) (.05)

Chi2 (p-value) 0.00 0.00 0.00 0.00

# of observations 118 131 131 131

R-Squared 0.60 0.53 0.69 0.69




45

Table 13 Impact of Sectoral Aid on Exports Method of Estimation 2SLS

(1)

2SLS

(2)

2SLS

(3)

2SLS

(4)


(.06)**


(.07)***


(.10)**


(.06)**

School Enrollment 0.01 0.02 -0.001 0.003

(.004) (.004) (.01)** (.01)

Exchange rate -0.12 -0.06 -0.13 -0.03

(.06)* (.06) (.10) (.08)

Economic policy -0.04 -0.08 -0.07 -0.05

(.06)** (.06)** (.09)** (.09)**


(.06)** (.05)*** (.07) (.06)**

Liberalization Policy -0.01 0.004 0.01 0.02

(.02) (.01) (.02) (.02)

GDP per capita -0.38 -0.20 -0.41 -0.04

(.14)*** (.15) (.15)*** (.21)

Chi2 (p-value) 0.00 0.00 0.00 0.00


R-Squared 0.16 0.43 0.01 0.02




46

6.0 Conclusion

In this paper we investigated some questions raised during the course

of this study. They are what component of aid is useful in promoting trade in

developing countries? We found that aid to sectors was useful in promoting

trade, with aid to trade policy, infrastructure, agriculture and education being

significant in promoting trade. This is consistent with past findings by

Morrissey et al (2004) and Velde and Cali (2009) who state that channeling

aid to productive sectors in an economy could boost export oriented growth.

Aid to industry had no impact on trade so we neglected it in our results. It

was likely that finished goods from developing countries do not compete

favorably with goods from developed countries and technology is often a

problem in developing countries making it difficult to produce. Effective aid

had contributed less to exports compared to official aid. However

multilateral aid proved more useful promoting export than bilateral aid this is

attributable to conditions associated with multilateral aid disbursements

which make them more effective in promoting exporting.

We also investigated if economic policies and institutional quality

improves or decreases aid effectiveness in promoting trade in Africa? We

found that economic policy and the quality of institutions in Africa generally

weakens the effectiveness of aid in promoting trade. The interactive

variables “aid*government economic policy” “aid*institutional quality” and

“aid*natural resources” had a reduced effect on trade. This is consistent with

past findings such as Burnside and Dollar (2002) and (2004) which state that

aid will be effective in the presence of good policies and other findings by

Sachs and Warner (1995) and Ross (2001) that suggest the presence of

natural resources and weak institutions can affect economic development in

developing countries. The inclusion of natural resources in our model caused

our model to suffer from misspecification resulting in negative R-squared so

we exclude it and used its interaction with aid in our subsequent GLS

regression. This interactive variable aid* natural resources exerted a reduced

effect on trade across all sectors, reducing aid effectiveness across sectors.

Therefore diversifying the economy in many African countries should

therefore be a strong concern to governments.

The policy implications of our findings is that economic policy has a

significant effect on aid effectiveness in Africa, therefore donors should

continue to emphasize the need for African countries to float sound and

consistent economic policies. Such policies could be vital in shoring up

investor’s confidence and ensure the effective use of aid to boost capacities

that can improve trade and stimulate export oriented growth on the long run.

Secondly channeling aid to sectors that are likely to improve export

capacities in developing countries could likely improve the way that aid can

be used to drive growth in an effective manner. Aid given to trade capacities


47

will likely fulfill the short term intention of giving aid to developing

countries since it is likely to contribute to export driven growth in many

African countries allowing for a discontinuation of aid giving policies to

promote growth. Over reliance on natural resources continue to remain an

impediment to the growth of other sectors in many African economies,

promoting diversification is likely to help prevent shocks (due to price

fluctuation in natural resources) in many African countries that rely on

specific natural resources for income. The reliance on these natural resources

as a source of alternative revenue often prevents governments from

implementing sound policies that could improve growth. Alternative revenue

sources through for example a creation of effective taxation scheme can help

create other sources of financing government activities thereby reducing

overdependence on resource derived revenues.

References:

Baltagi, B.H., 2005.Econometric Analysis of Panel Data.3rd edition. New

York: John Wiley and Sons.

Baltagi, B.H., and Wu, P.X., 1999. “Unequally Spaced Panel Data

Regressions with AR(1) Disturbances.” Econometric Theory, 15:814-823.

Boone, P., 1994. The impact of foreign aid on savings and growth. Mimeo,

London School of Economics.

Boone, P., 1996. “Politics and the effectiveness of foreign aid.”European

Economic Review, 40:289–329.

Bourguignon F. and Sundberg M., 2008 “Aid Effectiveness – Opening the

Black Box”.American Economic Review, 97(2) pp. 316 – 321

Burnside C. and Dollar D., 2000.“Aid, Policies and Growth”.American

Economic Review, 90 (4), pp. 847 - 868.

Cali, M. and D. W. Velde., 2011. “Does Aid For Trade Really Improve

Trade Performance?”,World Development, 39(5):725-740.

Central Intelligence Agency., 2011. CIA World Factbook, 2011.Accessed

online February 11, 2012 at http://cia.gov/library/publications/the-world-

factbook.

Chang, C. C.; Fernandez-Arias, E., and Serven, L., 1999.Measuring Aid

Flows: A New Approach. Washington: World Bank, Development

Economics Research Group.

Clemens, M.A.; C.J. Kenny & T.J. Moss., 2004. The trouble with the MDGs:

confronting expectations of aid and developments success. Working Papers

no. 40. Center for Global Development.

De Long, B. and B.Eichengreen.,1991. The Marshall Plan: History of the

Most Successful Structural Adjustment Program, NBER Working Paper

3899.

http://cia.gov/library/publications/the-world-factbook

http://cia.gov/library/publications/the-world-factbook

http://www.nber.org/papers/w3899

http://www.nber.org/papers/w3899


48

De Renzio, Paolo., 2006. Aids, Budgets and Accountability: A Survey

Article Development Policy Review, 24(6): 627-645.

Douclouliagos H. and Paldam M., 2007 “A meta-analysis of development aid

allocation: The effects ofIncome level and population size” University of

Arhus Denmark Working Paper.

Djankov et al., 2006. “Does Foreign Aid Help?” Cato Journal, 26 (1): 1-28.

Easterly W., 2003. “Can Foreign Aid Buy Growth?” Journal of Economic

Perspective 17 (3) pp. 23 – 48

EconomyWatch., 2010. Africa Economy, accessed online February 28, 2012

at http://www.economywatch.com/world_economy/africa/

Gormanee, K., Morrissey, O., Mosley, P. and A.Verschoor. 2003. “Aid,Pro-

Poor Government Spending, and Welfare.”CREDIT Research Paper.No

03/03.

Jempa, C.1991. The Tying of Aid, OECD Development Center, Paris Jensen,

T.M. 2008. How Does Aid Work? A Literature Review of Aid Effectiveness

and Welfare Studies.DIIS Working Paper, 2008/29.

Johnson, H. G., 1967. Economic Policies toward Less Developed

Countries.Washington: Brookings Institution.

Karingi, S.N. and V. Leyaro. 2009. Monitoring Aid for Trade in Africa: An

Assessment of the Effectiveness of the Aid for Trade. African Trade Policy

Centre, Working Paper No.83.

Knack S., 2001. “Aid Dependence And The Quality of Governance: Cross

Country Empirical Test”. Southern Economic Journal

McPherson, M.F. and T. Rakovski. 2001. “Understanding the Growth

Process in Sub-Saharan Africa: Some Empirical Estimates,” African

Economic Policy Discussion Paper.

Morrissey O., Osei R., McGillvray M., Lloyd T., 1992. “Investigating the

Relationship between Aid and Trade Flows”. A CREDIT (Centre for

Research in Economic Development and International Trade) Research

Paper, 38 (10), pp

Morrissey O., Osei R., Lloyd T., 2004. “The Nature of Aid and Trade

Relationships”. European Journal of Economic Research, 16 (2), pp. 354 -

374.

Mosley P., Hudson J., Vershoor A., 2004. “Aid Poverty Reduction and The

New Conditionality”. The Economic Journal, 114(496), pp. 217- 243.

Osborne E., 2002. “Rethinking Foreign aid” Cato Journal. 22(2): 297-316.

Pettersson J., 2004. “Foreign Sectoral Aid Fungibility, Growth, and Poverty

Reduction”.Journal of Economic Literature, 68(2), pp. 310- 329

Svensson J., 2000. “When Is Foreign Aid Policy Credible? Dependence and

Conditionality” Journal of Development Economics, 61(2000), pp. 61- 84.

Thirlwall, A. P. 1976. “When Is Trade More Valuable than Aid?” Journal of

Development Studies 13: 35–41.

http://www.economywatch.com/world_economy/africa/


49

United Nations, 2010.MDG Gap Task Force Report: The Global Partnership

for Development at a Critical Juncture, United Nations: New York.

Wooldridge, J.M. 2010.Econometric Analysis of Cross Section and Panel

Data, MIT Press: Cambridge, MA.

World Bank. 2012. World Development Indicators.Accessed online February

11, 2012 at http://data.worldbank.org.

Yeats, A. 1982. “Development Assistance: Trade vs. Aid and the Relative

Performance of Industrial Countries.” World Development 10: 863–69.

Appendix A. Table 14 Trade Regressions With Aid*Economic Policy

Interaction

Method of Estimation GLS

(1)

GLS

(2)

GLS

(3)

GLS

(4)


(.07)***

Aid to Trade*Policy 0.01

(.01)**

- - -

- - -

Aid to infrastructure - 0.16

(.19)***

Aid to Infrastruc.*policy - 0.01 - -

(0.01)**


(.13)***

Aid to Agriculture*policy - - 0.02 -

(.01)**


(.22)***

Aid to Education*policy - - - 0.01

(.01)**

Aid to industry - - - -

Aid to Industry*policy - - - -


(.003)*** (.003)*** (.003)*** (.003)***

Exchange rate 0.07 0.02 0.22 0.01

(.02)*** (.02) (0.14) (0.02)


(.04)*** (.04)*** (.04)*** (.04)***

Liberalization rate -0.03 -0.03 -0.03 -0.03

(.03)*** (.03)*** (.03)*** (.03)***

GDP per capita 0.07 0.08 0.07 0.08

(.05) (.05)** (.05) (.05)

Chi2 (p-value) 0.00 0.00 0.00 0.00


R-Squared 0.62 0.62 0.62 0.62



http://data.worldbank.org/


50

Appendix B. Table 15 Trade Regressions With Aid*Institutions Interaction Method of Estimation GLS

(1)

GLS

(2)

GLS

(3)

GLS

(4)


(.07)***

Aid to Trade*Institutions -0.04

(.01)***

- - -

- -

Aid to infrastructure - 0.84

(.28)***

Aid to Infrastr.*Institutions - -0.06 - -

(0.01)***


(.13)***

Aid to Agric*Institutions - - -0.04 -

(.01)***


(.22)***

Aid to Educ*Institutions - - - -0.04

(.01)**

Aid to industry - - - -

Aid to Industry*Institutions - - - -


(.003)*** (.003)*** (.003)*** (.003)***

Exchange rate 0.02 0.02 0.02 0.02

(.02) (.02) (.02) (.02)


(.03)*** (.03)*** (.03)*** (.03)***

GDP per capita 0.07 0.08 0.07 0.08

(.05) (.05)** (.05) (.05)

Chi2 (p-value) 0.00 0.00 0.00 0.00


R-Squared 0.62 0.62 0.62 0.62




51

Appendix C. Table 16 Trade Regressions With Aid*Natural Resources

Interaction Method of Estimation GLS

(1)

GLS

(2)

GLS

(3)

GLS

(4)


(.10)***

Aid to Trade*Resource 0.0004

(.0004)

- - -


(.40)***

Aid to Infrastr.* Resource - 0.0001 - -

(.001)


(.19)***

Aid to Agric.* Resource - - 0.0004 -

(.0005)


(.32)***

Aid to Educ.* Resource 0.0004

- - - (.0005)**

Aid to industry

Aid to Industry* Resource - - - -


(.004)** (.004)** (.004)** (.004)**

Exchange rate 0.04 0.04 0.04 0.04

(.03)*** (.03) (0.03) (0.03)


(.04)*** (.05)*** (.05)*** (.05)***


(.03)*** (.03)*** (.03)*** (.03)***

GDP per capita 0.07 0.08 0.07 0.07

(.07) (.07) (.07) (.07)

Chi2 (p-value) 0.00 0.00 0.00 0.00


R-Squared 0.61 0.61 0.61 0.61



Documents

DOES FOREIGN AID PROMOTE TRADE? EVIDENCE FROM SOME ... - Elpelpjournal.eu/wp-content/uploads/2018/03/1-1-2.pdf · to realize the MDGs by the 2015 deadline” therefore trade is vital